You are on page 1of 20

Valuing IT through Virtual Process Measurement

ICIS '94 Paper by:


Information and Operations Management University of Southern California, School of Business Administration Los Angeles, CA 90089-1421 E-mail: nissen@usc.edu; Tel 213-740-0196

Mark E. Nissen

Revised for ATRIUM 19 August 1994 The so called "productivity paradox" associated with information technology remains the focus of active research in information systems. One explanation involves the dearth of useful measures to assess the value of IT investments. A review of the predominant approaches to such measurement reveals a number of serious weaknesses and fundamental limitations. The research described in this paper addresses these limitations through a complementary methodology termed virtual process measurement (VPM). Through VPM, assessments of IT value are determined through the measurement of computer-based process representations (i.e., virtually), as opposed to measuring their real counterparts in ongoing organizations; this approach a ords a number of advantages that are unattainable through extant techniques. In this paper, the VPM methodology is discussed in considerable detail, and examples from industry practice are used to demonstrate the use and utility of this approach. The paper closes with a set of conclusions and some possible directions for continued research. Keywords Business process reengineering, Information systems, Knowledge-based systems, Process measurement, Qualitative reasoning, Valuation of information technology.
c 1994 Mark E. Nissen

Abstract

1 Productivity Paradox
A trillion dollars have been invested in information technology (IT) through the decade of the eighties, yet marked improvement in business productivity remains illusive 38]. This so called "productivity paradox" has been cited in the information systems (IS) literature as a lingering dilemma 4, 22], and represents the focus of current research in academics (e.g., 1, 14, 26, 46]). According to one IS journal editor, a key aspect to this dilemma pertains to the manner in which productivity is measured, exacerbated by the lack of good measures for the value of IT 27]. Given the deep and rapid proliferation of IT through the business environment, this need for useful IT-valuation measures will certainly grow through time; moreover, to the extent that measures for IT valuation fail to capture the "true" economic bene t of IT investments, business managers risk serious misallocation of capital to IT-based investments. The search for IT valuation measures is not new to IS research. Traditional nancial techniques (e.g., ROI, IRR) have been employed for decades, and measures have been developed to estimate IS functionality and cost (e.g., function point analysis, COCOMO). The key limitation of these approaches is that they fail to capture intangible costs and bene ts, many of which are inherently qualitative, very di cult to quantify, and nearly impossible to reduce to measures expressed in terms of dollars, complexity, and lines of code 16]. Other approaches take a strategic view of IT, discarding valuation measures for the promise of attaining competitive advantage; however, the sustainability of IT-based competitive advantages is arguable, so valuation remains key to analysis and decision making. Further, IS textbooks (e.g., 10]) now include guidance on subjective valuation, but this approach is asystematic, and su ers from inconsistency and bias. Many researchers now argue for methodological pluralism (e.g., 16, 24]), through which multiple simultaneous approaches are employed in an attempt to "triangulate" on the underlying IT value. In this paper, we too adopt this multiple-methodology approach, for the measures and approaches developed to date have considerable utility and merit; however, we strive to overcome the residual limitations of this ensemble of approaches through a methodology termed virtual process measurement (VPM). Virtual process measurement derives from a more-comprehensive methodology employed for business process reengineering. In this methodology, which is currently being actively and e ectively employed in a number of organizations, processes from the eld are represented in our knowledge-based modeling environment. This environment also supports symbolic simulation, in which dynamic process measures are taken from the represented process. A validation step ensures that the behavior of this representation is consistent with the real-world process in the eld, so that value-based measure- ments can be made regarding virtual processes; these include the obvious measures of process performance (e.g., cost and cycle time), in addition to a battery of measures with heuristic value that is useful for identifying process weaknesses and generating redesign alternatives. These measures are also useful for IT valuation, and can be used, for example, to compare the performance of an organizational process with the proposed IT investment, against that 1

Method Financial \ \ Functional \ Strategic \ Subjective \ \

Strengths Quantify tangible costs and bene ts Precise, accurate measurements Theoretical decision rules Assess costs early in SDLC Facilitate software engineering Focus upon competitive advantage Align IS with business strategy Focus upon lean, responsive process Focus upon managing risk Focus upon value added

Weaknesses Poor at qualitative concepts Poor at addressing intangibles Limited to \dollarized variables" Do not address bene ts Limited to software engineering costs Competitive advantage not sustainable Do not address fundamental economics How to measure? How to relate to IT investments?

Table 1: Key Strengths and Weaknesses of its counterpart without the IT implementation. Moreover, the rich, high- delity, dynamic process representation supports detailed measurement at the level of the organizational microprocess 32]; this unit of analysis provides a context-speci c view into the workings of alternative process designs, with particular emphasis on how IT implementations a ect the manner in which work is performed. In the next section, we discuss the predominant approaches toward IT valuation, and then describe virtual process measurement in considerable detail. Section four includes two examples of this methodology applied to an important process captured from the eld, and the paper closes with a set of conclusions and some possible directions for continued research.

2 IT Valuation
The predominant approaches to the valuation of IT can be classi ed in four major categories: 1) nancial, 2) functional, 3) strategic, and 4) subjective. Table 1 provides a summary of the key strengths and weaknesses of each to provide a backdrop for discussion of virtual process measurement. Financial measures can be e ectively employed to quantify tangible costs and bene ts associated with IT investments. These approaches can be found in standard nance textbooks (e.g., 45]), and also include the time-based measures employed in the practice of industrial engineering 3]. Use of nancial measures requires dollar-denominated measurement of costs and bene ts, generally in terms of cost and revenue streams over time. These approaches have the advantage of precision supported by quantitative measures, and their strong theoretical basis supports unambiguous decision rules regarding investment. Examples include net present value (NPV), internal rate of return (IRR), and payback period. The fundamental limitation of nancial measures derives from their (time- and) dollardenominated, quantitative nature: many important costs and bene ts associated with IT 2

2.1 Financial Measures

investments are qualitative, intangible, and di cult or impossible to measure quantitatively. Such limitation is very well recognized in the eld of arti cial intelligence (AI) 36], in which powerful symbolic representational techniques are employed to obviate this problem.

2.2 Functional Measures

Functional measures were developed in the context of information systems analysis and design, in which the objective was to estimate the complexity of an IS early in the systems development life cycle (SDLC). Such measures directly address IS costs associated with software engineering (SE). Examples include function point analysis 2, 8], which represents a systematic approach to estimating source lines of code (SLOC), the McCabe Measure of IS complexity, and COCOMO 4], which entails the use of a multivariate statistical model to estimate SE costs. Although these measures address SE costs, they do not address the important dimension of bene ts; this represents a fundamental limitation of methods in this class. Moreover, these functional measures focus only upon software engineering; hence, they ignore costs for hardware, and the potentially huge costs associated with implementation, training, and usage are left unaddressed. A number of publications have addressed IS strategy and strategic information systems 19, 23, 28, 35, 48, 49], the latter of which entail IS used to shape or enable a rm's business strategy; hence, strategic IS are described in terms of a unique class of importance 49, p. 18]. The key argument is that the precise costs and bene ts of strategic IS are relatively unimportant, certainly when the IS a ords a company the opportunity to dominate a competitive arena, or is required for a rm's survival. A number of consultants begin with the assumption that an IS is indispensable (and, hence, must be developed). Examples include using the competitive forces model 35] to change the nature of competition for a rm (e.g., by raising IT-enabled barriers to entry), or the theory of strategic thrusts 49] to outline IT-based means for achieving competitive advantage (e.g., through product di erentiation, innovation, or competitive alliance). However, cases like the computerized airline reservation systems 21] suggest that IT-enabled competitive advantages may not be sustainable. Once a competitive advantage has been mitigated, strategic IS lose some of their special status and importance, and the fundamental economics of cost vs. bene t and risk vs. return are re- imposed as key determinants of IT value. This Subjective Measures category represents something of a "catchall" for techniques not addressed in the categories above, many of which have been proposed in the trade literature. For example, the chief information o cer (CIO) of a major U.S. corporation adduced 12, 3

2.3 Strategic Measures

2.4 Subjective Measures

p. 35] "When IT is seen as part of the process of making organizations leaner and more responsive to the marketplace, the issue of placing a dollar value on systems starts to become irrelevant." In a related example, an IT consultant suggests that "managing risk" represents the key to e ective IT investment. Additionally, much has been written regarding the necessity of valuation within the rubric of Total Quality Management (TQM), with its emphasis on value added 11, 20]. However, a critical question remains without answer: How does one measure aspects such as "leanness," "responsiveness," "risk," "value added," and other qualitative concepts, in a manner that captures the contribution from investment in IT? Virtual process measurement provides an approach to measuring some of these concepts, and has been developed to help overcome the limitations inherent in the approaches summarized above. From the weaknesses identi ed in Table 1, we propose that VPM should minimally address seven measurement issues: 1) non-"dollarized" variables, 2) qualitative concepts, 3) intangibles, 4) IT-related bene ts as well as costs, 5) organizational costs above software engineering, 6) fundamental economics, and 7) a measurement procedure. We shall see that VPM addresses a number of issues beyond this minimal set.

3 Virtual Process Measurement


It is important to note that virtual process measurement has been developed as a complement to the ensemble of measurement approaches currently available. Hence the objective is to augment the portfolio of capabilities supported at present with a methodology that addresses the weaknesses inherent therein. Our discussion of VPM follows each of the three elements in the term: 1) virtual, 2) process, and 3) measurement. The term virtual is commonly used to describe phenomena or e ects that exist or occur in or through the computer. For example, the concept of the virtual organization 18, p. 67, 41] represents computer-supported communication, coordination, and processing su cient to enable an "organization" of workers to function, without the necessity of collocation or even belonging to a common organizational entity (e.g., rm, department, workgroup); thus the "organization" exists virtually, although its output could be indistinguishable from that of its corporeal counterpart. Similarly, the concept of the virtual marketplace 30] represents computer-supported information, communication, and commitment/transaction processing su cient to enable buyers and sellers to conduct commerce, without meeting personally or requiring a physical marketplace (e.g., store, market, stock exchange); thus the "marketplace" exists virtually, although the economic transactions may be indistinguishable from those of other market modes. In our present context, the concept of virtual IT valuation represents computer-supported representations, simulations, and visualizations of the e ects produced by IT in the organizational context (e.g., decreased cost, improved quality, shorter cycle time), without actually introducing the IT into the processes of an ongoing organizational concern (e.g., corporation, 4

3.1 Virtual

government agency, military unit); thus the "IT valuation" is assessed virtually, although the measurements used for assessment could be indistinguishable from those obtained from ongoing organizations in the eld. The key to virtual IT valuation is consistency between the behavior of a computer-based process representation and its ongoing organizational counterpart in the eld. Representational validation is central to AI, through which a number of e ective techniques have been developed. For example logic can be employed to verify that the represented process is internally consistent and possible, given the set of organizational constraints corresponding to the process in the eld. Additionally, the representation can be validated against its real-world counterpart by ensuring that the number of organizational agents, their roles, task assignments, tools, coordinative structures, work environment, and like attributes match. Further, simulation can be employed to validate the dynamic behavior of the represented process{for example, to ensure that the time required for agents to perform tasks, the number and types of agents required to perform these tasks, the throughput and quality of products/services through the process, and like attributes are consistent between the real process and its virtual counterpart. Finally, visualization can be employed to provide process experts with the means to witness the mechanics of represented processes, and with su cient delity to make detailed comparisons with observable process mechanics in the eld. A number of advantages accrue from virtual IT valuation. For example, the operations of an ongoing organization do not need to be interrupted while the IT-valuation e ort is being conducted. Additionally, a wide diversity of (good and bad) processes can be examined virtually, without the requirement to physically implement or pilot the process changes in the eld organization. Further, because of the time compression enabled by simulation 40], a very large number of alternative process designs can be observed and evaluated in a relatively short period of time (i.e., much shorter than the cycle time required for the process in the eld). Moreover, process operations can be replicated again and again via computer, whereas each "live" process instance is unique in some regard (e.g., time sequence, exogenous factors, measurement error). Finally, related to replication, experimental controls can be imposed upon the study of a virtual process: through computer support, the experimenter can selectively modify one variable at a time, as well as variables in any combination, to observe and measure the e ects on performance of the represented organization. Together these advantages enable IT valuation imbued with high internal validity (through experimental controls) while simultaneously preserving the external validity of the represented process. No existing valuation approach o ers such important advantages. The process represents a key element of VPM. With this approach, the value of IT is assessed by measuring the impact of a proposed investment in IT on the a ected organizational processes. The term process has been de ned 18, p. 35] ". . . as a collection of activities that takes one or more kinds of input and creates an output that is of value to the customer." Viewing a process systems-theoretically 44], it can also be described in terms of inputs, 5

3.2 Process

outputs, a boundary, transfer function, and mission or goal. This facilitates a number of representational economies, for nearly every organizational process can be represented via this systems scheme, and "families" of reusable process models can be organized through object-oriented meta-models 31]. Processes are necessarily cross-functional 17] and dynamic 5, p. 6], so they provide a rich representation of organizational operations; that is, not only do they describe who (e.g., organization- al agent/role) does what (e.g., organizational activities/tasks), but also how (e.g., with which kinds of tools and communication modes) this work is accomplished, and when the various agents, tasks, and tools come together to add value. Additionally, the process represents the central unit of analysis for business process reengineering (BPR) 15], for which IT represents a key enabling technology 5, p. 37]. Additionally, the systems-process scheme supports high- delity computer representation 13], because very detailed process models can be represented through attributed directed graphs (A- digraphs); the A-digraph constitutes the fundamental data structure employed in support of VPM, for, not only can it be used to link process inputs and outputs through value creation, but it can also be used to capture the rich organizational details associated with the process task environment. Using the A-digraph for process representation a ords several key advantages. First, many important qualitative and intangible concepts associated with modeled processes can be represented symbolically, de ned with rigor 33], and analyzed through qualitative simulation 47]. Additionally, graph theory provides a powerful set of mathematics to support systematic analysis and reliable measurement of organizational processes. Further, embedding the A-digraph process representation within a knowledge-based system (KBS) enables measurements and analyses to be performed automatically; not only does this support the VPM methodology, but it also represents a step toward reengineering the process of reengineering itself. The direct or fundamental measures have been developed through the procedure of extension 37]; this enables measures to be de ned with su cient rigor, and measurements to be obtained with su cient precision, to support the use of ratio scales in our measurement methodology. Ratio scales are very desirable in terms of their information content and inferential power, but are generally quite di cult to obtain in the social sciences 29]. Process measures can be conveniently classi ed according to three categories: static, dynamic, and improvement. Static attributes include direct measures such as length, breadth, IT- based processes/communications, organizational hando s, and others, in addition to a number of derived metrics such as parallelism, bushiness, IT-processing/communication density, and Value Chain fraction, among many others. These measures have the advantage of being domain independent, in that any process can be described in terms of such metrics, and compared with other (even cross- industry) processes. Similarly, dynamic attributes can be de ned and measured directly through computer 6

3.3 Measurement

simulation. For example, the cost, cycle time, throughput, quality, and other dynamic constructs can be simulated and measured at any level of detail in a represented process. Such measurements support performance comparisons between alternative process designs, which can include, for example, the speci c incorporation of an IT investment under consideration. A number of improvement attributes are being developed for incorporation into VPM. One use of improvement attributes would be to represent the time-based derivatives of their dynamic counterparts (e.g., cycle time improvement, cost reduction, risk mitigation). Quantum improvement measures have been employed to gage the degree to which a reengineering proposal is "radical" 18] (e.g., 100 50] can be used to capture an important dimension of organizational learning 9]; additionally, improvement attributes such as process adaptability, exibility, and repairability provide the opportunity to represent an organization's potential for improvement, deemed to be important in today's competitive environment 42]. Work on improvement measures is just beginning, but the static and dynamic measures are relatively well-established, and play a key role in the VPM methodology developed and employed to date. To summarize, the approach of virtual process measurement includes a methodology for the representation, simulation, and visualization of organizational processes, in terms of static, dynamic, and improvement attributes that are domain-independent, and which can be measured with rigor and precision. This methodology approaches the task of IT valuation through comparison: the performance of a validated, high- delity model of a real process in the eld is measured with respect to any number of modi ed processes that re ect implementation of various IT investments under consideration. Di erences are attributed to the IT investments, and IT value is assessed on the basis of organizational process performance. As such, VPM can be used to complement other available IT-valuation methods. Table 2 provides a summary of the weaknesses identi ed above for current valuation methods, and the manner in which VPM addresses their limitations. Provided that the structure and behavior of an organizational process model is consistent with that of the represented process in the eld (consistency which is ensured through model validation), VPM can be employed to e ectively address the de ciencies identi ed above for current IT- valuation methods. This approach not only extends our capabilities for valuing IT, but it provides a rich, high- delity model of the organization and its processes, which can be used for reengineering, training, process documentation, and a number of other purposes.

3.4 Summary

4 Measurement Example
This example is taken from the order ful llment process of a major U.S. corporation, and focuses upon the speci c sub-process of credit approval, which represents a key element of order ful llment for many companies; a similar case is discussed in 18, pp. 36-39] in a reengineering context, which serves to highlight the close coupling of IT valuation and 7

Issue Non-"dollarized" variables Qualitative concepts \ Intangibles \ \ IT-related bene ts \ Costs above software \ Fundamental economics \ \ Measurement procedure \ \

VPM Approach Model variables not limited to dollars Symbolic modeling supports qualitative concepts Qualitative simulation projects behavior of qualitative characteristics Qualitative variables can represent intangible costs and bene ts AI techniques enable inference regarding intangibles Detailed process models can be visualized for intangible aspects Multidimensional performance bene ts are projected through simulation Bene ts like process simplicity and improvement can be assessed Process models address agents, organizations, tasks, tools and communications related to process performance, not just software Costs and bene ts can be measured Elements of risk and return can be assessed VPM output can support analysis by other valuation methods VPM obtains measurements from process models Process models are validated against processes in the eld Process measurements are calibrated against eld processes

Table 2: Issues Addressed through VPM reengineering. However, the purpose of this example is to explicate VPM, not to discuss an exemplar of "radical" reengineering. Figure 1 presents a rich pictures 6] view of the baseline credit approval process. A eld sales representative with a new-customer lead contacts the credit approval organization by telephone to request a credit proposal for the potential customer; the relevant customer and sale information is written-down by a contact person in the credit organization, who then carries the information to the Credit Department. Four primary departments participate in the credit approval process: Credit, Practices, Pricing, and Quote; agents from each department perform the credit check, terms development, payment calculation, and proposal preparation activities serially. From the rich pictures representation, notice the isolated presence and usage of IT in each department, and note that all communication is accomplished via telephone and paper; notice also that a third party is contracted for delivery. In this baseline case, work within each department must be approved by the department manager before being sent to the subsequent department for processing; this model captures the "command and control" behavior exhibited in many organizations. The rst step required for virtual process measurement is to represent a process via A-digraph. A level-1 schema is presented in Figure 2. This schema indicates that the baseline level-1 pro- cess is comprised of ve tasks, which are represented as process-activity nodes connected by task- precedence edges. Attached to this digraph (but not shown) are a number 8

4.1 The Credit Approval Process

4.2 Representation

Field Sales Credit Dept.

Practices Dept. Contact Delivery

Quote Dept.

Pricing Dept.

Figure 1: Rich Pictures Representation - Baseline

Figure 2: Articulator Representation - Baseline Level-1 of relevant attributes asso- ciated with this process, including the organization (credit approval in this case), role (agent specialty or job description), processing mode (tools used for accomplishing work), communication mode (method of communication), and Value Chain. In Figure 3, we present the sequence of level-2 activities for the quote task from above. Notice the feedback loop which captures an aspect of output quality: unless the quoteletter re ects acceptable quality, this task chain must be repeated until quality meets the modeled threshold. Elements of quality are determined during the modeling process, and quality checks are performed dynamically during simulation.

Table 3 presents descriptions of some static measures that can be applied to the baseline credit approval process, along with the direct measurements that were obtained from the level-2 process schemata. These direct measurements provide some summary information regarding the baseline process, and also provide the basis for a set of derived measures, some of which are listed in Table 4. The derived measurements provide additional static information, and are useful to generate process redesign heuristics; a selection of process measurements and heuristics generated for the baseline process is presented in Table 5 and discussed below. First, the parallelism measurement of 1.00 indicates that the process is entirely sequential; 10

4.3.1 Measurements

4.3 Measurements and Redesign

Figure 3: Articulator Representation - Baseline Level-2


Measure Steps Length Leaves Levels Breadth Hando s IT Activities (IT-A) IT Communications (IT-C) Value Chains (VCs) Calculation Base Total number of nodes 24 Nodes in longest path 24 Number of atomic activities 18 Levels of decomposition 2 Distinct paths 1 Organizational traverses 6 Processing steps supported by IT 7 Communications supported by IT 0 Value Chain traverses 3

Table 3: Static Process Measures (Direct)


Measure Parallelism Bushiness Hando Fraction IT-A Fraction IT-C Fraction VC Fraction Footprint Calculation Steps divided by length Breadth divided by length Hando s normalized by steps IT-A normalized by steps IT-C normalized by steps VCs normalized by steps Levels multiplied by leaves Base 1.00 0.04 0.25 0.29 0.00 0.13 36

Table 4: Static Process Measures (Derived) 11

Measure Value Redesign Heuristic Parallelism 1.00 Parallel processing of tasks Hando Fraction 0.25 Case manager/team or empowerment IT-A Fraction 0.29 IT tools to support tasks IT-C Fraction 0.00 IT tools to support communication VC Fraction 0.13 Forward or backward integration

Table 5: Measurement-Based Redesign Heuristics heuristically, this suggests the possibility of recon guring the work ow so that one or more of these steps are processed in parallel, as opposed to serially. Second, the hando fraction mea- surement of 0.25 indicates that the process is highly-departmentalized; heuristically, this suggests the possibility of employing a case manager 18, p. 36] or case team 5, p. 97] approach, or empowerment can also be instituted to decentralize and accelerate decision making 43]. Third, the values for IT-A fraction and IT-C fraction indicate that the IT density of the process is relatively low; heuristically, this suggests that the one or more process activities and communications may be better accomplished with IT support. Finally, the measured Value Chain fraction indicates considerable complexity in the Value Stream, as three separate Value Chains are involved in this relatively small process; heuristically, this suggests that forward or backward integration may be used to eliminate one or more Value Chain interfaces. For our rst process redesign alternative, we hypothesized that three of the level- 1 activities could be performed in parallel, so the tasks assigned to specialists in Credit, Practices, and Pricing are conducted concurrently. Further, an IT investment is made to couple the separate IT systems through an integrated database, and enable work to ow electronically through the four departments. Additionally, some empowerment is introduced so that work can ow between departments without managerial approval, provided that it meets quality requirements; this re ects a process redesign that does not require changes to the organization chart. We refer to this alternative as "Redesign-1." For our second process redesign alternative, which we refer to as "Redesign-2," we extended the process design from above, by making an additional investment in IT-supported pro- cessing and communication in the eld. This enables the eld sales representative to contact the credit organization electronically, as well as view the status of proposal requests as they move through the process. Electronic communication also enables forward integration of the delivery task, as electronic communication obviates the need for a third party. Figure 4 illustrates the rich pictures representation of Redesign-2 for comparison with the baseline, and Figure 5 serves to delineate the redesigned (level-1) work ow con guration; notice that the process design is now visibly shorter.

4.3.2 Redesign

12

Field Sales

Credit Dept.

Integrated Database Quote Dept. Practices Dept.

Pricing Dept.

Figure 4: Rich Pictures Representation - Baseline

13

Figure 5: Articulator Representation - Redesign-2 Level-1 To determine the value of these IT investments, we need to compare the process before IT implementation with its post-redesign counterparts. Table 6 provides a comparison of some key static measurements obtained from the three process designs. The reader can see that both redesigned processes are smaller, shorter, and involve fewer steps (i.e., simpler); Redesign-2 also re ects the forward integration of a Value Chain. Notice that both redesigned processes re ect greater parallelism, and have much higher IT densities in terms of both activities and communica- tions. Results such as these (e.g., process size, simplicity, forward integration, parallelism, IT den- sity) re ect some of the intangible bene ts deriving from the IT investments; other, more tangible bene ts like dynamic performance improvements were projected and measured through simulation. Table 7 provides a comparison of some key dynamic measurements obtained from the three process designs. In the qualitative simulations, token resource units were projected through the modeled process to measure various performance categories, including cost, cycle time, throughput, and risk. Qualitative simulation is powerful, because it supports dynamic process measurement with minimal information, and it can be conducted at a very early stage of process understanding and redesign. However, unlike the ratio scales supported by the static measures above, qualitative simulations used to obtain these dynamic measurements support only ordinal relations between process design alternatives; although ordinal relations are not as powerful as their ratio counterparts 37], they provide for useful comparisons and interesting results. 14

4.4 IT Valuation

Measure Baseline Redesign-1 Redesign-2 Steps 24 18 17 Length 24 14 13 Footprint 36 24 22 Breadth 1 3 3 Value Chains 3 3 2 Parallelism 1.00 1.29 1.31 IT-A Fraction 0.29 0.72 0.88 IT-C Fraction 0.00 0.44 0.59

Table 6: Comparative Static Measures


Measure Baseline Redesign-1 Redesign-2 Agent Cost 18 12 11 IT Cost 9 21 25 Overhead Cost 27 6 5 Supplier Cost 1 1 0 Cycle Time 27 17 15 Communications Risk 19 17 5 Throughput 9 15 17

Table 7: Comparative Dynamic Measures One interesting result can be seen regarding the tradeo between the costs of agents ("Agent Cost") and those associated with investments in IT ("IT Cost"); this tradeo represents the known economic phenomenon of labor-capital substitution 7], and its simulation signi es the kind of high- delity process models supporting VPM. Also, di erences in (allocated) overhead cost re- ect the shorter cycle times and higher throughput rates of the redesigned processes; shorter process cycle times decrease period costs per cycle, and higher throughput rates increase the volume of output used for allocation; this result is consistent with measurements obtained through most accounting systems in practice. Another interesting result pertains to communications risk. This measure captures the management heuristics that telephone communications are more error-prone, and paperbased disseminations are harder to keep current, than those conducted through electronic media. Notice that the hybrid of electronic communications and telephone/paper-based systems in Redesign-1 barely a ected communications risk, even though they contributed toward substantial improvements in cost, cycle time, and output; simulation of this riskreturn relationship indicates another aspect of the IT-valuation problem that is di cult to explore through existing valuation methods. Many additional measures, results, and comparisons would be required to use VPM for IT valuation in practice, but the IT investments assessed in this section should su ce to convey the use and utility of VPM, along with its potential to complement extant IT-valuation methods. Table 8 summarizes a few quantum improvement measurements obtained from these examples; the "Redesign 2M" column re ects marginal improvements over Redesign-1. 15

Measure Baseline Redesign-1 Redesign-2 Total Cost 27 % 25 % -3% Cycle Time 37 % 44 % 12 % Communications Risk 11 % 74 % 88%

Table 8: Quantum Improvement Measures

5 Conclusions
Through our discussion of virtual process measurement, and the examples above, a number of conclusions can be drawn, and several opportunities for future research emerge. First, using VPM, we were able to analyze and compare a number of alternative process designs, each re ecting a di erent level of investment in IT. We were able to assess the relative, multidimensional value of IT across the various designs, in a manner that overcame the limitations of current valuation methods. This satis es the primary objective of the present research e ort. However, continued research will be required to determine the relative e cacy of various measurement constructs, to extend static and dynamic measures to capture domain-speci c factors, to de ne and test improvement metrics, and to validate that virtual process measures are consistent with measures obtained directly from ongoing processes in the eld. Second, we were able to see how IT valuation and VPM are closely coupled with reengineering. For example, our static measurements were employed to conceive of the IT investments and redesign alternatives examined above. This measurement-based approach to redesign has the potential to introduce a considerable amount of engineering into the process of "re- engineering." Additional research can be conducted to develop a theoretical framework for process redesign, say in terms of a process-transformation model, and to test such a model in both laboratory and eld environments. Research oriented toward the automation of the redesign process is already underway in the laboratory, but the use and evaluation of this knowledge- based engineering methodology in practice will be indispensable. Finally, the measures and examples from above have an apparent internal focus, as they are oriented primarily toward e ciency as opposed to e ectiveness; however, nothing associated with the VPM methodology, KBS, or other support tools precludes the modeling and simulation of competitive arenas, for example, which could be employed to measure and assess the e ects of alternative IT investments and strategies. Basic research into issues such as the representation- al ontology, units of analysis, and epistemology associated with such use of VPM is only now being contemplated. These questions suggest a number of opportunities for continued research along the lines discussed in this paper.

6 Acknowledgements
I wish to thank Dr. Malcolm Munro, ICIS '94 Program Co-chair and the reviewers for their helpful comments on an earlier draft of this paper, and to acknowledge the guidance and 16

support from Dr. Walter Scacchi, who directs the USC ATRIUM laboratory where this present research was conducted.

7 References
1] Abdel-Hamid, T., "Software Project Control: An Experimental Investigation of Judgement with Fallible Information" IEEE Transactions on Software Engineering 19, No. 6 (June 1993), pp. 603- 612. 2] Albrecht, A. and J. Ga ney, "Software Function, Source Lines of Code, and Development E ort Prediction," IEEE Transactions on Software Engineering SE-9, No. 6 (1983), pp. 639-647. 3] Barnes, R.M., Motion and Time Study: Design and Measurement of Work New York, NY: Wiley (1980). 4] Boehm, B., Software Engineering Economics Englewood Cli s, NJ: Prentice-Hall (1981). 5] Davenport, T.H., Process Innovation: Reengineering Work through Information Technology Boston, MA: Harvard Press (1993). 6] Davies, L. and P. Ledington, Information in Action: Soft Systems Methodology Hong Kong: Macmillan (1991). 7] Douglas, E.J., Managerial Economics: Theory, Practice, and Problems (Second Edition) Englewood Cli s, NJ: Prentice-Hall (1983). 8] Dreger, J., Function Point Analysis Englewood Cli s, NJ: Prentice-Hall (1989). 9] Epple, D., Argote, L. and R. Devadas, "Organizational Learning Curves: A Method for Investigating Intra-Plant Transfer of Knowledge Acquired through Learning by Doing" Organization Science 2, No. 1 (1991). 10] Fertuk, L., Systems Analysis and Design: With CASE Tools Dubuque, IA: W.C. Brown Publishers (1992). 11] Flood, R.L., Beyond TQM New York, NY: Wiley (1993). 12] Freedman, D., "The ROI Polloi" CIO (April 1990), pp. 30-40. 13] Gasser, L, Hulthage, I., Leverich, B., Lieb, J. and A. Majchrzak, "Organizations as Complex, Dynamic Design Problems" in: M. Figueiras and L. Damas (Eds.) Progress in Arti cial Intelligence Lecture Notes in Arti cial Intelligence 727 Portugal: Springer Verlag (1993). 14] Glazer, R., "Measuring the Value of Information: The Information-Intensive Organization" IBM Systems Journal 32, No. 1 (1993), pp. 99-110. 15] Hall, G., Rosenthal, J. and J. Wade, "How to Make Reengineering Really Work" Harvard Business Review (November-December 1993), pp. 119-131. 16] Hall, R., "The Strategic Analysis of Intangible Resources" Strategic Management Journal 13 (1992), pp. 135-144. 17] Hammer, M., "Reengineering Work: Don't Automate, Obliterate" Harvard Business Review (July-August 1990), pp. 104-112. 18] Hammer, M. and J. Champy, Reengineering the Corporation: A Manifesto for Busi17

ness Revolution New York, NY: Harper Business Press (1993). 19] Henderson, J.C. and N. Venkatraman, "Strategic Alignment: Leveraging Information Technology for Transforming Organizations" IBM Systems Journal 32, No. 1 (1993), pp. 4-16. 20] Ho herr, G.D., Moran, J.D. and G. Nadler, Breakthrough Thinking in Total Quality Management Englewood Cli s, NJ: Prentice-Hall (1994). 21] Hopper, M.D., "Rattling SABRE{New Ways to Compete on Information" Harvard Business Review (May-June 1990), pp. 118-125. 22] Hu , S.L., "Call for Papers" 1994 International Conference on Information Systems (1994). 23] Kambil, A., Henderson, J.C. and H. Mohsenzadeh, "Strategic Management of Information Technology Investments: An Options Perspective" to appear in: R. Banker, R. Kau man, and M. Mahmood (Eds.) Strategic Information Technology Management: Perspectives on Organizational Growth and Competitive Advantage Middletown, PA: Idea Group Publishing (1993), pp. 161-178. 24] Kanevsky, V. and T.J. Housel, "Value-Based Business Process Reengineering: An Objective Approach to Value Added" Manuscript (1993). 25] Kaplan, R.S. and D.P. Norton, "Putting the Balanced Scorecard to Work" Harvard Business Review (September-October 1993), pp. 134-147. 26] Keen, P.G.W., "Information Technology and the Management Di erence: A Fusion Map" IBM Systems Journal 32, No. 1 (1993), pp. 17-39. 27] King, J.L., "Unraveling the Productivity Paradox" University of Southern California Presentation (1992). 28] Konsynski, B.R., "Strategic Control in the Extended Enterprise" IBM Systems Journal 32, No. 1 (1993), pp. 111-142. 29] Krantz, D.H., Luce, R.D. Suppes, P. and A. Tversky, Foundations of Measurement, Volume I: Additive and Polynomial Representations New York, NY: Academic Press (1971). 30] Malone, T.W., Yates, J. and R.I. Benjamin, "Electronic Markets and Electronic Hierarchies" Communications of the ACM 30, No. 6 (June 1987), pp. 484-497. 31] Mi, P. and W. Scacchi, "A Uni ed Model of Software Systems, Agents, Tools, and Processes" ATRIUM Working Paper, University of Southern California, School of Business Administration (1993). 32] Nissen, M., "Qualitative Simulation of Organizational Microprocesses" in: J. Nunamaker, Jr. and R. Sprague, Jr. (Eds.) Proceedings of the Twenty-Sixth Hawaii International Conference on System Sciences (Volume IV) (1994), pp. 635-644. 33] Nissen, M., Knowledge-Based Organizational Process Engineering Diss. in progress, University of Southern California, School of Business Administration (1994). 34] O'Guin, M.C., The Complete Guide to Activity-Based Costing Englewood Cli s, NJ: Prentice- Hall (1991). 35] Porter, M.E. and V.E. Millar, "How Information Gives You Competitive Advantage" Harvard Business Review (July-August 1985), pp. 149-160. 36] Rich, E. and K. Knight, Arti cial Intelligence (Second Edition) New York, NY:

18

McGraw-Hill (1991). 37] Roberts, F.S., Measurement Theory: with Applications to Decisionmaking, Utility, and the Social Sciences Reading, MA: Addison-Wesley (1979). 38] Rothschild, M., "The Coming Productivity Surge" Forbes ASAP (29 March 1993), pp. 17-18. 39] Scacchi, W., "Modeling, Enacting, and Integration Complex Organizational Processes" ATRIUM Working Paper, University of Southern California, School of Business Administration (1993). 40] Senge, P.M., The Fifth Discipline: The Art and Practice of The Learning Organization New York, NY: Bantam Doubleday Dell (1990). 41] Schroth, R. and C. Mui, "Focusing on the Future: Advanced Technology Groups" Indications 8, No. 2 (November/December 1991). 42] Stalk, G., Jr. and T.M. Hout, Competing Against Time: How Time-Based Competition is Reshaping Global Markets New York, NY: Free Press (1990). 43] Stohr, E.A. and B.R. Konsynski (Eds.) Information Systems and Decision Processes Los Alamitos: IEEE Press (1992). 44] von Bertalan y, L., General System Theory: Foundations, Development, Applications New York, NY: George Brazillier (1968). 45] Welsch, G.A. and R.N. Anthony, Fundamentals of Financial Accounting (Revised Edition) Homewood, IL: Irwin (1977). 46] Weill, P., "The Relationship Between Investment in Information Technology and Firm Performance: A Study of the Valve Manufacturing Sector" Information Systems Research 3, No. 4 (1992), pp. 307-333. 47] Weld, D. S. and J. de Kleer (Eds.) Qualitative Reasoning about Physical Systems San Mateo, CA: Morgan Kaufmann (1990). 48] Westland, J.C., "The Marginal Analysis of Strategic Investments in Information Technology" to appear in: R. Banker, R. Kau man, and M. Mahmood (Eds.) Strategic Information Technology Management: Perspectives on Organizational Growth and Competitive Advantage Middletown, PA: Idea Group Publishing (1993), pp. 161-178. 49] Wiseman, C., Strategic Information Systems Homewood, IL: Irwin (1988). 50] Wright, T.P., "Factors A ecting the Cost of Airplanes" Journal of the Aeronautical Sciences 3 (1936), pp. 122-128.

19